Hey all, just wanted to to post on my recent difficulty 10 win with spire. Some additional thoughts from my difficulty 9 run previously and some info on parts of the game that were abused to get the win. I've also included an archive with all 30 or so saves that detail my journey through this run, to support some of my points here, and in case it helps any of you.
Link:
https://drive.google.com/file/d/1KnhH4nJ0MeX4Hr8A0CYmmujiP-0z4RQF/view?usp=sharing(Please let me know if this link doesn't work for any of you)
First off, a victory screen
Victory was eventually via Spire ship power allowing me to bully the AI through to their homeworld. I didn't bother with imperial fleet or going above mark 5 city because it would have been the same result with some more waiting. *shrug*
Strategic-Level Stuff1. Starting map seed RNG: Simply, and especially at this difficulty, getting bad RNG with, *especially*, the types of fleets and ARS you get within 2/3 jumps of your starting point is extremely important. It can be possible to basically lose a difficulty 10 game at this point with bad RNG.
My strong opinion is that the hard random RNG we have in the game right now with ARS and fleets should be completely removed. Even tier 2 planets nearby require teching up your starting fleet to clean them out at a decent speed, without losing too much metal, and so you don't have the luxury of sitting on tier 1/2 tech... ie, if your fleets and ARS don't match up with what you have chosen as your initially strategy, it's much better to just randomize the map again and restart. This sort of RNG is a waste of time.
Check the save entitled "2 Parasites secure" to see what my opening map/moves looked like - I got an excellent parasite/high DPS generalist combo with my gunbots right off the bat.
I understand that some on the dev team want to force random selections so that a player needs to "roll with the punches" and figure out how to win with what they are given, but at high difficulties, there is no margin for error, and in my opinion, this is just bad game design, since it gives some players a very easy experience (RNG works out in their favor), and some players a hard experience, and additionally does not allow them to play the way they like. At difficulty 10, this becomes "converts the game from winnable, to unwinnable, if your RNG sucks". Basically, starcraft does not roll a dice every game to decide what faction you are going to play as - in the same vein, AIW2 shouldn't roll a dice deciding for me what ships I can use.
2. Starting Fleet: Selection here is incredibly important. I cannot state how weak some of the fleets are, (the classic fleet comes to mind), and how absolutely neccessary it is to start with the cloak fleet. At difficulty 10, having a strong cloak element is *mandatory*, so starting the game with a cloaked transport and locking that in is very important.
In addition, at difficulty 9+, speed is absolutely everything. Every minute you spend destroying an AI structure, neutering something, etc.. hurts you, because it gives more AI budget to reinforce planets, warden, etc.. etc..
Regarding the support fleets, the engineering battlestation and rejuvenator support fleet's repair abilities *vastly* overpower the rest of the selectable ships. In the early game, the speed that you can reinforce your fleet is absolutely crucial to clearing out early worlds fast, and against a difficult 10 AI, you will find yourself abusing repair on shields to keep your homeworld alive quite often. It seems like significant balance work needs to be done across the board between the starting strike/support and battlestations.
3. Primary Ship Strategy: Following with the starting fleet, there seems to be not much flexibility in what ships can be used in a difficulty 10 run. Again, you are forced into high DPS, and the fact that the AI is not able to intercept ships effectively at the tactical level, and is quite vulnerable to micro'd ship kiting, just makes it worse. 100 vanguards will do absolutely nothing (outside of extremely niche challenges) for you, taking eons to kill anything, and not even pulling priority off of high-priority ships like the botnet golem. On the other hand, 100 vanguards worth of gunbots can destroy entire AI planets by themselves if properly microed (I speak more about this down in the tactical section).
You can check my fleet choices and deep strikes looking at the "5 Rushing before CPA" save where I knew the CPA was coming up and was forced to dive deep for a crucial botnet golem.
4. Officer / Lone fleet balance is all over the place: I think some testing and balance time needs to go into the officers/lone fleets, as currently, many of them aren't really worth the AIP costs. At this difficulty level, every AIP counts, as well as every piece of metal, so deciding what to take is very important. In this run, I found that pretty much all officers besides the widow golem and botnet golem were worthless, with the hive golem being "acceptable", though I didn't bother getting it. Taking a planet's worth of AIP in addition to the officer/lone cost is incredibly damaging in the early/mid game... at difficulty 10 i've noticed a huge difference even just between 35 and 50 AIP.
In contrast to the huge AIP penalties of taking these, looking for fleet expanders and ARSs is magnitudes more beneficial.
5. Office / Lone Fleets are annoying to micro: On the more opinion side of things, since most of the high value officers/lone ships are so crucial, expensive, and have such a high targeting priority, they must be *constantly* micro'd, or they end up at 1% hp, which is basically game ending (5 million metal down the drain is a game over, basically). I'm not sure if this is a good thing, intended, or whatnot, but I really enjoy focusing on strategic choices, rather than excessive micro, so having to babysit 2+ ships like this in different places all the time was annoying. *shrug*
6. Scourge usage: Ai reaction to Scourge is pretty strong, even when they are very weak, and you will see Exo units come in quite early. Due to this, I use scourge mainly as a force multiplier, *after* I have a solid chokepoint between the AI homeworld and safe parts of the galaxy. If you check the save "14 let the bullying commence" you will see the initial steps taken using the scourge. Generally, the scourge as they are right now are useful for two things:
1. Easily handling small AI fleets that sneak through chokepoints, to avoid the frustration of chasing AI raiders around all day.
2. Exploiting their grow-over-time potential, while sitting in your chokepoint for 3+ hours (I didn't do this, because it's boring)
Though I didn't use it in this playthrough, allowing the scourge to grow while holding a major chokepoint for a few hours, and then pushing out is probably the most optimum strategy right now. For their cost/AI reaction, the scourge are *incredibly* overpowered right now. My opinion is that scourge/spire not being very tied to AIP is a huge mistake, and the current ability to sit and do nothing for hours while your scourge maxes out their growth is something that, imo, should have been caught in beta.
The scourge are quite good at following in on a player's fleet, so if you look into the save "18 SW push for scourge", you will see how I have set everything up to clean up the SW quadrant of the map. (actually in this run I didn't exploit scourge to the max, and just pushed out on my own for the most part - honestly, the scourge never really did anything "amazing" for me, as I never waited a few hours for them to grow, as I felt that the balance on this part of the scourge is that bad - the way I did it in this run feels more like what they were "intended" to be balanced to be... by themselves, not very powerful, but a small helpful force multiplier with a bit of a drawback)
7. Spire usage: I went into this in depth in my last post, so a lot of my feedback is located there. To sum up though, the spire capital city and it's defenses were crucial for me to lock down the center of the map, and the way that the spire relic-chase-based gameplay is designed doesn't seem to be very challenging. Divorcing spire AI reaction from AIP is wrong, in my opinion, and the current AIP and base AI difficulty should weigh *much* heavier on the response that the AI gives you.
Besides that, my suggestions on changing the spire to be more of an "open war" with the AI opening up wormholes and such across the galaxy makes a lot more sense than the relic farming/chokepoint abuse that currently exists. If you go to my later saves, such as "25 City 5 good spawn", you will see how the RNG nature of the relic spawn process can make the "ramp up" from spire city 1 to 12 extremely abusable, easy and boring. I was able to keep a large part of the galaxy 2+ jumps away from my planets, and by having my main choke also very close to the AI homeworld (feature of this map), the algorithm is forced to give me a ~70%+ chance of getting relics popping basically on exactly the systems I want them to spawn anyways.
Even without this abuse, by using my cloaked fleet to pop (and immediately delete) relics that spawned near the AI homeworld, I would eventually get the cities I wanted. The entire idea of spawning relics randomly throughout the galaxy is flawed, imo, as all it does is reward chokepoint/abusive playstyles.
Finally, and I had hoped this would change at difficulty 10, but it did not, is the issue of the AI not ramping up it's response to spire appropriately. Currently, the way spire is designed, once you get past 5 cities or so, and assuming you have a decent chokepoint, it is difficult to lose the game, and a win is almost inevitable. If a modification of Spire towards an open war, as per my suggestion, is not in the cards, then at the least, someone experienced at playing the game with spire needs to measure their progress at difficulty 9/10 after starting the spire campaign, and the algorithm/difficulty needs to be tweaked to offer similar difficulty to the early game CPAs that are so difficult to deal with at this difficulty level. The challenge faced after starting the spire campaign at difficulty 9 needs to be defeating experienced players most of the time, and on difficulty 10, the AI should be vaporizing most players, if the intention of difficulty 10 is a difficulty that most players should not be able to beat.
8. Zombie Farming: Currently, zombies, for whatever reason, do not have a very high chance of attacking things ~2 or more jumps away from where they are spawned. This makes them *excellent* defenders, even with the default zombie settings which are supposed to make them go out and attack stuff. What this means, and you will see it by looking through saves 18+, is that you can very effectively get large amounts of zombies "patrolling" your safe spaces, as long as you farm them in areas where they will not be destroyed by overwhelming forces.
This leads to one of the major ways I survived the massive 2-hour mark CPA, and that is through timing a long 120 second hack through the central chokepoint system (it had a coordinator on it) to finish right before the CPA started. Luckily, and again this comes down to RNG somewhat, I had a triple parasite fleet stack (very rare, by the way), allowing me to get a pretty good AI->zombie conversion rate going as long as I was able to kill AI ships fast enough. This 120 second hack, in addition to how powerful the hack responses are at difficulty 10, allowed me to farm a few hundred strength of zombies in the exact system that the CPA was forced to go through. This is one of the main reasons I was able to get through the 2 hour CPA.
My suggestion is that zombies need to be a bit more aggressive in attacking AI worlds, instead of sitting around in safe space all day. Regarding hacks, my opinion is that, at higher difficulties, the hack itself should respect what the AI is doing at that time - by that I mean, if the AI is charging up a CPA or exo-galactic wave, instead of loads of units showing up at the hack point, instead, that strength should be added directly to the CPA/exo wave, perhaps with a multiplier to compensate for the fact that that strength is going to be coming through a chokepoint instead of behind the chokepoint, which is usually the case when a player is doing a hack. This is very punishing to the player, though, so I would suggest it is added only to difficulty 9+, or even reserved for difficulty 10 only.
9. Only specific technologies are useful: Specifically, if you are going for scourge/spire, half the tech in the tech tree becomes useless. For example, a mark 7 military station on a chokepoint is *insanely* powerful when combined with the spire. Meanwhile, the same tech points spent on "citadels" will give you the strength of a single spire cruiser or two... the difference in effectiveness of different techs in the tree is mind-boggling. The only techs I found worthwhile (and extremely worthwhile they were), were the core weapon techs for my early/mid-game fleet, military command center, and engineering. Everything else was basically a waste of science. My opinion is that techs need to be balanced considering at least a difficulty 9 playthrough, and especially considering most people will want to use spire/scourge/etc... I guess in a difficulty 5 playthrough, you can feel good about your omega-powerful citadels "crushing" the AI, but at the highest levels of challenge, you can't throw away science like that.
Tactical-Level Stuff1. Abuse of manual micro and defeat in detail: Extremely powerful, but fragile, ships are required at this difficulty. By manually microing them to stick together, and stay just at the edge of the range of the leading "tendril" of incoming enemy ships, you can vaporize small parts of an enemy force, as long as you are *at least* a similar speed to them, or faster. Check the savegame entitled "Going into CPA with bot" - this was the first CPA of the game, and by far the hardest challenge through this entire run. I lost at least 20-30 times from this save before finally getting through it, and using this form of micro with my primary hotkey 1 fleet, along with the botnet golem, was the only way I was able to beat it.
The fact that every second counts against you means that this high DPS, high micro strategy is by far the most viable for beating a difficulty 10 AI.
Sidenote: AI worlds full of counter sniper turrets and velocity-based damage with infinite range crush your entire fleet when following this strategy. Better pray to the RNG gods that your key worlds are accessible without going through a world like this... though in this play-through I did actually have to bypass two planets to get to the botnet quickly enough, which was.. interesting to say the least.
2. Wormhole jump abuse: The AI does not react very intelligently to fleets jumping in/out of wormholes back and forth very quickly, and I found that one of the most successful ways to keep my high-value ships alive (mainly the botnet and widow), was to keep my engineering fleet sitting on the safe side of a wormhole, and jump my fleets back and forth in/out. This is the main reason why the widow golem is decent for it's AIP cost, as pulling 100+ enemy ships through the wormhole to be killed is very effective, especially on hard systems that otherwise would be nearly impossible to clear out.
3. Repair and forcefield abuse: On many occasions, I was forced to repair forcefields surrounding my home stations to keep the game going. By cycling the forcefields in and out of the enemy ship's line of fire, you can get engineers repairing them at a pretty quick rate, and especially with very large incoming shots, if they break the shield, the rest of the shot's damage disappears into thin air, which is extremely critical when you end up with a random artillery golem or the like trying to snipe your command center. AI ship targeting priorities are also not always locked onto the home station, and rotating forcefields like this seems to help mess with their targeting, sometimes causing them to target other ships, buildings, or forcefields instead of the home station/command station.
4. Standard cloak abuse: I spoke a bit about this in the strategic part of things, but basically, proper micro and usage of a cloaked transport ship (or raid frigates, which work pretty well also), is mandatory. There are too many critical objectives that you cannot easily get to without cloak, such as data centers and the like. I'm not sure if cloak should be so heavily required like this, since if you don't start with a cloaked fleet in the starting menu, you could get screwed on RNG.... In any case, as it is now, cloak micro is mandatory, and is actually not very difficult, as long as you have Mk4 cloak or so to work with. (watch out for forcefield bubbles though....)
Sidenote: 1% hull flagships should not be locked out of wormholes by forcefield bubbles. It is impossible to delete them and get them back, so this "feature" needs to be removed asap. I had to reload the game many times due to a random dire forcefield guardian covering key wormholes that my deep strike fleet needed to come back on, resulting in game over, basically.
5. AI Ship targeting priority abuse: Some specific ships in the game are able to take priority, depending on what ships the AI are using, away from *any* target *INCLUDING* the home command station. In this way, the botnet golem, widow golem, and things like decoy drones out of a few support fleets allowed me to play with the AI and pull them off of things I wanted to repair/defend (usually forcefields and command stations). I call this abuse because a rational actor that had, for instance, 500+ strength on a system, should focus-fire the command center *regardless* of whether a botnet golem is firing at them or not. In addition, the difference between a mark 7 military station staying up for 10 seconds, and blowing up, are immense... far greater than the actual metal value of the decoy drones/etc.. we are talking about.
Considering that the AI only gets more powerful over time, the AI targeting algorithm should be tweaked to ignore *ALL* other targets, and snipe down command centers (especially home command). Without this, decoy drones, botnet golems, widow golem, and similar, are just too powerful compared to trash like pike corvettes, lone spire frigates, etc... Balance changes down the line to command station defenses, forcefield HP, and similar will also need to be made... if the functionality of decoys and similar being able to divert enemy fire from command stations *is* desired, however, it makes sense that the mark level of these ships should influence how effective they are - perhaps mark 7 AI ships would have a 25% chance of ignoring mark 5 drones, for example.
In conclusion: This difficulty 10 run was very enjoyable, up until spire city 4+, where the game was already won, and I was sad that the AI didn't have the tools neccessary to organize an effective response. (I really would like to be able to enjoy fighting tooth and nail against motherships and the like). Overall though, a wonderful, unique experience that definitely offers something that doesn't exist in any other game.
